Temporal profiles of high-mobility group box 1 expression levels after cortical spreading depression in mice

Abstract

Introduction

Cortical spreading depression (CSD) has recently been shown to induce the release of the nuclear protein termed high-mobility group box 1 from neurons, causing activation of the trigeminovascular system. Here, we explored the effects of single and multiple cortical spreading depression inductions on high-mobility group box 1 (HMGB1) transcriptional activity relative to high-mobility group box 1 protein expression levels and intracellular localization in cortical neurons and astrocytes.

Methods

Single or multiple cortical spreading depression inductions were achieved by KCl application to the mouse cerebral cortex. The animals were sacrificed at 30 minutes, 3 hours and 24 hours after cortical spreading depression induction. High-mobility group box 1 expression levels were explored with in situ hybridization, Western blotting and immunostaining.

Results

Cortical spreading depression up-regulated high-mobility group box 1 transcriptional activity in neurons at 3 hours in a manner that was dependent on the number of cortical spreading depression inductions. At 24 hours, the high-mobility group box 1 transcriptional activity had returned to basal levels. Cortical spreading depression induced a reduction in high-mobility group box 1 protein expression at 3 hours, which was also dependent on the number of cortical spreading depression inductions. Following cortical spreading depression, the release of high-mobility group box 1 from the nucleus was observed in a small proportion of neurons, but not in astrocytes.

Conclusion

Cortical spreading depression induced translocation of high-mobility group box 1 from neuronal nuclei, driving transcriptional up-regulation of high-mobility group box 1 to maintain protein levels.

Introduction

In 25% of migraine cases, a transient focal neurological disturbance, termed an aura, precedes or accompanies the headache (1). Cortical spreading depression (CSD) is a slow propagating wave in the cerebral cortex that spreads at 2–5 mm/minute with depolarization of neurons and glia and is accompanied by secondary cerebral blood flow (CBF) changes. The phenomenon was first described in rabbits by Leao (2). Comparison of aura-associated visual symptoms and the propagation speed of CSD-related CBF changes led to the idea that CSD is the underlying mechanism of migraine auras (3). This was substantiated by a functional magnetic resonance imaging study demonstrating that blood oxygenation level-dependent signal changes advance contiguously over the occipital cortex in a manner that may explain the patient’s visual perception (4). From the pharmacological viewpoint, a preclinical study indicated that the electrical stimulation threshold for induction of CSD in the rat cortex increases after chronic treatment with five different migraine prophylactic drugs (5). In addition, the gap junction inhibitor tonabersat inhibits CSD and, as expected, decreases the occurrence of auras in migraineurs (6). Collectively, CSD is thought to be the neurophysiological correlate of auras. CSD is also observed in patients with cerebrovascular diseases such as subarachnoid haemorrhage (SAH) and malignant stroke (7). Using electrocorticography, Dreier et al. (8) reported that spreading depolarization, which likely represents CSD, was observed in 13 of 18 patients with SAH. Such spreading depolarization occurs in clusters and is responsible for delayed ischaemic brain damage after SAH. Dohmen et al. (9) conducted a study in 16 patients with large middle cerebral artery infarction and CSD was detected in all but two patients. Most depolarization events occurred in temporal clusters. Such CSD clustering may be seen in a subpopulation of migraineurs. Iizuka et al. (10) observed very long-lasting (up to 4 days after the onset of aura symptoms) CBF changes that were consistent with CSD occurrence with multiple sources during prolonged auras in patients with familial hemiplegic migraine type 2 (FHM2). In such cases, CSD can be harmful to brain tissue due to the considerable bioenergetics burden on tissue (11) and/or activation of biochemical changes detrimental to brain tissue (12).

High-mobility group box 1 (HMGB1), a ubiquitously expressed non-histone DNA binding nuclear protein, stabilizes the nucleosome structure (13) and regulates gene transcription (14). Upon injury, HMGB1 translocates to the cytoplasm and the extracellular space and serves as a damage-associated molecular pattern or alarmin (15). Moreover, several lines of evidence show that HMGB1 is released from neurons deprived of blood flow following middle cerebral artery occlusion (16–19). As treatment with an HMGB1-neutralizing antibody significantly ameliorates brain ischaemic injury (16–18), HMGB1 plays a harmful role in the pathophysiology of brain ischaemia. Remarkably, a recent study by Karatas et al. (20) has implicated an inflammatory mediator in the pathophysiology of CSD and migraine. According to their hypothesis, HMGB1 released from neurons acts on astrocytes comprising the glia limitans and induces prolonged synthesis of inflammatory mediators, which somehow induces nociceptive stimulation of the dural trigeminal terminals. Hence, HMGB1 may be a critical molecule that links the occurrence of auras (CSD) and the generation of migraine headaches (activation of the trigeminovascular system).

In the present study, we primarily explored the temporal profile of HMGB1 de novo synthesis relative to HMGB1 protein expression in the cerebral cortex after CSD induction. We also compared the effects of single and multiple CSD inductions on alterations in HMGB1 dynamics.

Materials and methods

Animals

Male C57BL/6 mice (CLEA Japan Inc., n = 42, aged 8–10 weeks, 22.0–25.2 g) were used in this study. They were housed in cages with free access to water and food. Eighteen animals were used for in situ hybridization and immunohistochemistry and the remaining animals were used for Western blot analysis. All experimental procedures were approved by the Animal Welfare Committee of Keio University (Authorization No. 09182-1).

Mice were randomly divided into six groups. The control group did not undergo any surgical procedures. In the other groups, after skin incision, a craniotomy was performed using a dental drill (Tas-35XL, Shofu, Kyoto, Japan) with the dura intact for 1 week before CSD induction to minimize the obligatory effect of surgical tissue injury. This was based on our preliminary data showing that HMGB1 transcription was up-regulated immediately after craniotomy and that the up-regulation subsided after 1 week (data not shown). CSD was induced once or five times. The mice subjected to a single CSD induction were sacrificed 3 hours after CSD (CSD1x-3 h group). Those subjected to five CSD inductions were sacrificed at 30 minutes, 3 hours or 24 hours after the final CSD induction (CSD5x-30 m, CSD5x-3 h, CSD5x-24 h groups). Sham-operated mice underwent craniotomy and were sacrificed 1 week later.

CSD induction

Mice were anaesthetized with isoflurane (1.0% in room air with a flow rate of 400 mL/minute), which proved not to affect the occurrence of CSD (21). All procedures were performed at 37℃. To measure direct current (DC) potentials, an Ag/AgCl electrode (tip diameter=200 µm, EEG-5002Ag; Bioresearch Center Co., Nagoya, Japan) was inserted above the pia mater (2 mm posterior and 2 mm lateral to bregma). Ag/AgCl reference electrodes (EER-5004Ag; Bioresearch Center Co.) were placed in the subcutaneous tissue ipsilateral to the CSD induction side. DC potentials were amplified at 1–100 Hz and digitized at 1 kHz with a differential head stage and differential extracellular amplifier (Models 4002 and EX1; Dagan Co., Minneapolis, MN, USA). At the site of craniotomy, we installed a small open cranial window with a diameter of 1.5 mm at 4 mm posterior and 2 mm lateral to bregma. After the cerebral cortex was exposed, a cup-shaped plastic tube was inserted into the cranial window and CSD was induced by applying 1 M KCl solution to the cortical surface. The cup-shaped plastic tube was used to prevent diffusion of the KCl solution to the dura mater from causing a nociceptive effect. The induction of CSD was verified by the appearance of typical deflections of DC potentials. We regulated the number of CSD inductions by washing out the KCl solution with physiological saline at an appropriate time point. Five CSD inductions could be achieved within 30 minutes.

Tissue preparation

Under deep anaesthesia by excess halothane (Fluothane; Takeda Pharmaceutical Company, Osaka, Japan), the animals were transcardially perfused with 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.0. Immediately after the perfusion fixation, the brain was dissected out and immersed in the same fixative for 4 hours at 4℃ and was then kept in 0.01 M phosphate-buffered saline (PBS) solution containing 30% sucrose (w/v) for cryoprotection. The parietal cortex located approximately 1.5 mm away from the site of CSD induction and 2.5 mm posterior to the bregma was studied. Subsequently, the brains were embedded in Tissue TEK (Sakura Finetek, Torrance, CA, USA) and were frozen in liquid nitrogen. Serial sections of 10 µm thickness were prepared on a cryostat (Leica CM 3050S; Leica Biosystems, Nussloch, Germany) in the horizontal plane along the long axis (22). Serial sections were used for immunostaining and in situ hybridization.

Immunohistochemistry

The sections were pre-incubated with 10% normal donkey serum/0.1 M phosphate buffer for 30 minutes for blocking. They were incubated with specific primary antibodies for 24 hours at room temperature. After they were washed with 0.01 M PBS, the sections were incubated with species-specific fluorophore-labelled secondary antibodies for 2 hours at room temperature. After rinsing with 0.01 M PBS, the sections were coverslipped in mounting medium (buffered glycerol: pH 8.6). The primary antibodies and dilutions used in the procedures were as follows: mouse anti-NeuN antibodies (code MAB 377; Millipore, Billerica, MA, USA; 1:100); mouse anti-GS6 antibodies (code MAB 302; Millipore; 1:200); rabbit anti-HMGB1 antibodies (code ab18256; Abcam, Cambridge, MA, USA; 1:1000). The epitope for the HMGB1 antibody used in this study is shown in Figure 1. Immunoreactivity was visualized using species-specific donkey secondary antibodies that were conjugated to Cy3 or fluorescein isothiocyanate; all secondary antibodies were obtained from Jackson Immunoresearch Laboratories (West Grove, PA, USA). Nuclear staining was performed with 4’,6-diamidino-2-phenylinodole (DAPI). The immunolabelled specimens were examined under a Keyence BIOREVO BZ-9000 microscope (Keyence, Osaka, Japan) and a TCS-SP5 confocal laser scanning microscope (Leica Microsystems, Mannheim, Germany).

Figure 1.

Schematic representation illustrating the locations of sense and anti-sense probes for in situ hybridization and the antibody epitope for detecting HMGB1. Grey arrows show designed riboprobes that correspond to nucleotides 12–643 of the HMGB1 cDNA. Red bar shows the epitope for the anti-HMGB1 antibody (Abcam) used for immunohistochemistry and Western blot analysis. The antibody epitope was residue 150 to the C-terminus of human HMGB1.

In situ hybridization

The GeneBank database was searched for mouse cDNAs encoding HMGB1 (BC064790.1). Using the DIG Labeling Kit (SP6/T7; Roche, Penzberg, Germany), we produced digoxigenin (DIG)-labelled anti-sense and sense riboprobes that corresponded to nucleotides 12–643 of the HMGB1 cDNA (Figure 1). Tissue sections were treated with proteinase K/PBS (10 µg/ml) for 5 minutes at room temperature. Subsequently, they were pre-hybridized for 1 hour at 65℃ and then hybridized with the DIG-labelled probes at 65℃ overnight. After washing, alkaline phosphatase-coupled anti-DIG fragment antibody (1:1000) was incubated at room temperature for 2 hours. Antibody binding was detected with the alkaline phosphatase reaction with nitro-blue-tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate. As a negative control, the sense probe was used in place of the anti-sense probe under the same experimental conditions. Microscopic analysis was performed with Keyence BIOREVO BZ-9000. Within the parietal cortex subjected to CSD, 10 regions of interest with an area of 10,000 µm² were assigned with five regions each in the superficial and deep layers. The average number of cells per mm² in these regions of interest was designated as the representative density value for the animal. Three animals were examined for each time point.

Western blot analysis

We dissected the parietal cortex subjected to CSD and the tissue was quickly rinsed with 0.01 M PBS. Subsequently, the dissected brain cortices were homogenized in ice-cold lysis buffer (RIPA buffer, 1×Complete (Roche), 1×PhoSTOP (Roche)). The resultant tissue lysates were subjected to sodium dodecyl sulfate polyacrylamide electrophoresis (SDS-PAGE). The separated proteins were electrically transferred to polyvinyl difluoride membranes (22). The primary antibodies and dilutions used in the study were as follows: anti-HMGB1 antibodies (1:1000); rabbit anti-glyceraldehyde 3-phosphate dehydrogenease (GAPDH) antibodies (code 2118; Cell Signaling Technology, Danvers, MA, USA, 1:1000). The immunoreactive bands were visualized using enhanced chemiluminescence and detected using a luminoimage analyser (LAS-4000; Fujifilm, Tokyo, Japan). Densitometric analysis of immunoreactive bands for HMGB1 and the internal control (GAPDH) was performed using the image analysis software Multi Gauge version 3.0 (Fujifilm).

Statistical analysis

The data are shown as the means±SD. Data were evaluated with one-way analysis of variance followed by Tukey’s correction for multiple comparisons. We used IBM SPSS, ver. 22 (Chicago, IL, USA) for statistical analysis. Statistical significance was set at p < 0.05.

Results

In situ hybridization of HMGB1 mRNA

As depicted in Figure 2(a,b), in situ hybridization analysis (n = 3 in each group) revealed the presence of HMGB1 mRNA-positive cells in the cerebral cortex of control animals with weak signal intensity in each cell (control: 463.7 ± 36.7/mm²). Morphologically, the signal was detected in neurons and smaller cells that appeared to be glial cells (Figure 2(c), left image, arrow). In the sham group, we observed a similar density of HMGB1 mRNA-positive cells (sham: 452.0 ± 105.0/mm², p = 1.00 vs. control). At 3 hours after single CSD induction, the mean density of HMGB1 mRNA-positive cells in CSD1x-3 h was 769.0 ± 379.5/mm² (p = 0.51 vs. control). Regarding multiple CSD inductions (five), we found a significant increase in HMGB1 mRNA-positive cells 3 hours after CSD compared to the control group (CSD5x-3 h: 1400.0 ± 139.8/mm², p < 0.01 vs. control). Concomitantly, the mRNA signal intensity was enhanced in each positive cell (Figure 2(a)). Although such CSD-induced alterations in HMGB1 transcriptional activity were obvious in neurons, no discernible change occurred in glia-appearing small cells (Figure 2(c), right image, arrow). As we did not observe any significant change in the number of HMGB1 mRNA-positive cells 30 minutes after multiple CSD inductions (CSD5x-30 m: 948.0 ± 281.5/mm², p = 0.12 vs. control), we deduced that the HMGB1 transcriptional activity was gradually up-regulated. Twenty-four hours after multiple CSD inductions, the number of HMGB1 mRNA-positive cells had returned to the basal level (CSD5x-24 h: 645.7 ± 103.2/mm², p = 0.89 vs. control) with diminished signal intensity in each cell.

Figure 2.

In situ hybridization for HMGB1 mRNA. (a) Representative images using the anti-sense and sense probes in each group are presented. The upper and lower columns show anti-sense and sense images, respectively. Scale bar, 100 µm for all images. (b) The densities of HMGB1 mRNA-positive cells (cells/mm²) are shown in a bar graph (mean±SD). We observed a significant increase in HMGB1 mRNA-positive cells in the CSD5x-3 h group (**p < 0.01 vs. control). (c) High-power images are shown. Arrows indicate cells smaller than neurons, which appear to be glial cells. In the control group, the signal intensity of these cells was similar to that of neurons. In the CSD5x-3 h group, despite an increased signal intensity in neurons, we observed no significant change in signal intensity in glia-like cells. Scale bar, 10 µm for all images.

Alterations in the number of HMGB1-positive cells and HMGB1 intracellular localization after CSD induction

At baseline, HMGB1 immunostaining (n = 3 in each group) combined with nuclear counterstaining with DAPI revealed that HMGB1 was chiefly localized within the nucleus. Double immunostaining for HMGB1 and either the neuronal marker NeuN or the astroglial marker GS6 indicated that HMGB1 was expressed in both cell types (Figure 3(a)), which was consistent with the in situ hybridization data. Immunostaining disclosed that the density of HMGB1-positive cells decreased at 3 hours after multiple CSD inductions (Figure 3(b)). Quantitatively, HMGB1 expression level revealed by Western blot analysis (n = 4 in each group, Figure 3(c,d) in CSD1x-3 h was 0.61 ± 0.27 compared to the baseline level that was set at 1.0 (p = 0.09 vs. control). Meanwhile, there was a significant reduction after multiple CSD inductions (CSD5x-3 h: 0.38 ± 0.15, p < 0.01 vs. control). However, even with multiple CSD inductions, we did not observe any significant change at 30 minutes (CSD5x-30 m: 0.88 ± 0.20, p = 0.94 vs. control). Twenty-four hours after multiple CSD inductions, the HMGB1 protein level returned to baseline (CSD5x-24 h: 0.85 ± 0.24, p = 0.88 vs. control). Almost no change was seen in the sham group (sham: 0.93 ± 0.12, p = 1.00 vs. control).

Figure 3.

Immunohistochemistry and Western blot analysis for HMGB1. (a) Double immunohistochemical staining for HMGB1 and either NeuN (neurons) or GS6 (astrocytes) in the control group is shown. HMGB1 immunoreactivity was observed in the nucleus of NeuN-positive cells (left) and GS6-positive cells (right). Scale bar, 10 µm for all images. (b) A representative image of immunostaining for HMGB1 in each group after CSD is shown. Scale bar, 100 µm for all images. (c) Western blot analysis of HMGB1 and GAPDH (internal control) is shown. (d) The ratios of the HMGB1/GAPDH signal intensity are given in a bar graph. A significant decrease was observed in the CSD5x-3 h group (**p < 0.01 vs. control). The ratio returned to baseline in the CSD5x-24 h group.

Regarding intracellular HMGB1 localization, extranuclear immunoreactivity was seen in a small proportion of NeuN-positive neurons in the cerebral cortex following CSD. Morphologically, the HMGB1 immunostaining pattern suggested release into the cytoplasm from the nucleus. In contrast, such HMGB1 intracellular translocation was not observed in GS6-positive astrocytes (Figure 4).

Figure 4.

Intracellular HMGB1 localization after CSD induction. Double immunostaining for HMGB1 and either NeuN or GS6 in the CSD5x-3 h group is shown. Extranuclear immunoreactivity was observed in a small proportion of NeuN-positive cells in the cerebral cortex following CSD (arrow). No such change was seen in GS6-positive cells (arrowhead). Scale bar, 10 µm for all images.

Discussion

In the present study, we showed the temporal profile of HMGB1 expression at the RNA and protein levels in the cerebral cortex following CSD.

Our in situ hybridization study detected only a minimal signal intensity in the cerebral cortex of control and sham-operated mice despite the overt HMGB1 protein expression as revealed by immunostaining and Western blot analysis. From these findings, we infer that de novo HMGB1 synthesis in the cerebral cortex is nearly inactive with a low turnover rate in the basal condition.

Employing double fluorescent immunostaining, we identified HMGB1 protein immunoreactivity in both neurons and astrocytes in the cerebral cortex. The immunoreactivity was localized in the nucleus in basal conditions. Consistent with the report by Karatas et al. (20), HMGB1 immunoreactivity was also observed in the cytoplasm in a small proportion of neurons after CSD, supporting the notion that CSD promotes the translocation of HMGB1 from the nucleus. In contrast, the intracellular localization of HMGB1 was not altered in astrocytes. To the best of our knowledge, such a differential effect of CSD on HMGB1 distribution is novel, although a similar observation has been reported in the ischaemic brain (16). The role of astrocytes in CSD-related pathophysiology is controversial. Astrocytes play an inhibitory role in the development and maintenance of CSD by buffering extracellular potassium and glutamate (23). The gene that causes FHM2 when mutated, ATP1A2, encodes the α2 subunit of the glial Na,K-ATPase, which is important for the regulation of extracellular potassium and glutamate concentrations (24). A knockin mouse heterozygous for the FHM2 causative loss-of-function mutation (ATP1a2^+/R887) exhibits increased susceptibility to CSD, providing in vivo evidence for a critical role for astrocytes in CSD development (25). Meanwhile, some studies have shown that astrocytes are primary drivers of CSD or regulators of CSD-induced changes in the vascular calibre (26,27). Despite such implications for astrocytes in various aspects of CSD, our findings indicate that HMGB1 release is not likely a CSD-relevant pathophysiological event in astrocytes.

Our Western blot analysis revealed a significant reduction in HMGB1 protein level only 3 hours after multiple CSD inductions. This decrease in the HMGB1 protein level was clearly dependent on the number of CSD inductions, because a single CSD induction did not cause any significant change in HMGB1 protein expression in the identical time frame. Given the results of our in situ hybridization study, the reduction was clearly not due to decreased transcriptional activity. On the contrary, we observed a significant up-regulation of HMGB1 transcriptional activity 3 hours after CSD. The protein expression level had normalized 24 hours after CSD induction. Considering such a temporal profile, we interpret the enhanced transcriptional activity as a compensatory mechanism by which cortical neurons replenish the normal reserve of HMGB1 protein. We presume that the down-regulation of transcriptional activity between 3 and 24 hours after CSD occurred after completion of HMGB1 replenishment. Moreover, in conjunction with the above-mentioned extra-nuclear HMGB1 immunoreactivity in neurons after CSD, the reduced level of HMGB1 appeared to be due to its release into the extra-cerebral tissue. HMGB1 release from neurons has been reported in rodent stroke models (28,29) and is consistent with clinical data demonstrating secretion into the blood stream (17,30). Moreover, in our immunohistochemical analysis, HMGB1 immunoreactivity was observed in the nucleus of remaining HMGB1-positive neurons, while Karatas et al. (20) demonstrated that a considerable proportion of HMGB1 immunoreactivity was present in the cytoplasm immediately after CSD. The changes detected in tissue HMGB1 levels at 30 minutes after CSD were not significant, which is again not consistent with the data by Karatas et al. (20). These disparities may be due to the difference in the methods for inducing CSD. A salient finding in the time course of brain HMGB1 expression after CSD was that a shorter duration was required for the recovery of the protein level compared to cerebral ischaemia, in which normalization of the HMGB1 protein level was not observed until several days after ischaemic insult (31). Such a discrepancy may be explained by a difference in the effect of the insults on cell viability, which is closely related to cellular transcriptional and translational activity. The dependency of the CSD-induced dynamics of HMGB1 expression on the number of CSD inductions may be clinically relevant. In terms of the duration and symptomatology, only a single CSD appears to be induced in usual cases of migraine aura. However, patients with FHM may display a lowered threshold for CSD induction (25). A neuroimaging study on patients with FHM2 revealed CBF changes indicative of multiple CSD occurrences (10). Furthermore, using electrocorticography, clusters of spreading depolarization with persistent depression, which likely reflects the occurrence of CSD, have been demonstrated in patients with SAH and malignant stroke (8,9). Hence, HMGB1 may play a greater role in the pathophysiological mechanisms of these disorders compared to usual migraine episodes.

Our observation that single CSD did not cause a significant change in HMGB1 level appears to be contradictory to the assertion that HMGB1 release is relevant to migraine headache, which is likely to occur after single CSD (20). However, since we observed a decreasing trend of HMGB1 at 3 hours after single CSD, it is inferred that a minor release of HMGB1 from neurons is occurring. Hence, we cannot exclude the possibility that such a small change in HMGB1 can trigger a mechanism that ultimately activates the trigeminovascular system.

If HMGB1 release and transcriptional induction are acting as a trigger for migraine headache, as asserted by Karatas et al. (20), why do not all patients with ischaemic, haemorrhagic or traumatic injury experience migrainous headache? This enigma may be explained as follows. It is well known that migraine attacks can be pharmacologically induced in migraineurs. For example, nitric oxide (NO) donors, like nitroglycerin and glyceryl trinitrate, are known to trigger migraine attacks. They can cause typical migraine headache almost exclusively in migraineurs several hours after administration (32). Meanwhile, NO donors induce only a transient headache presumably attributable to cranial vasodilatation in non-migraineurs. Moreover, Hansen et al. (33) observed migraine-like attacks in 8 out of 14 migraineurs when exposed to calcitonin gene-related peptide, whereas no migraine-like attacks were observed in 11 healthy controls. These observations emphasize that there exists an intrinsic mechanism inherent to migraineurs, which can be easily activated by the above-mentioned pharmacological measures. We deduce that the HMGB1 release from neurons can be connected to the disease process that leads up to the development of migraine headache only in migraineurs.

A big strength in our study is that we analysed de novo synthesis of the HMGB1 molecule by performing in situ hybridization. Consequently, we first discovered that HMGB1 is a highly inducible protein after CSD induction. Moreover, our immunostaining data along with those by Karatas et al. (20) revealed that HMGB1 is abundant in neuronal and glial nuclei in the basal condition. This has a clinical relevance. For example, judging from these findings, it is prudent to use a strategy for inhibiting HMGB1 transcription as well as pre-formed HMGB1 protein, if we were to thoroughly inhibit HMGB1 function after CSD. On the other hand, we should cautiously consider the possibility that a difference in the durations of 1 M KCl application may have affected the extent of HMGB1 release from neurons. In our experiment, a longer period of exposure time was required to induce CSD five times than a single time. Such longer KCl exposure may have caused tissue injury and, as a consequence, a more marked HMGB1 release may have been induced. However, as we observed a subsequent up-regulation of transcriptional activity in neurons, it is obvious that neuronal integrity was not lost completely. Another caveat of our study is that we have yet to clarify whether the extracellular release and de novo synthesis of HMGB1 after CSD have a detrimental or beneficial effect on the brain. This point should be addressed in future studies.

In summary, our data support the previous finding that HMGB1 release from neurons is induced by CSD (20). In addition, our study provides novel information about the temporal profile of HMGB1 mRNA transcriptional activity in relation to HMGB1 protein expression after CSD, the differential effects of CSD on the expression level and intracellular distribution of HMGB1 between neurons and astrocytes, and the dependency of such alterations on the number of CSD inductions.